Infrared spectra of argon matrix-isolated alkali ... - ACS Publications

Ahmed M. Sadoon , Gautam Sarma , Ethan M. Cunningham , Jon Tandy , Magnus W. D. Hanson-Heine , Nicholas A. .... Bridgett E. Coleman and Bruce S. Ault...
0 downloads 0 Views 1MB Size
2426

Journal of the American Chemical Society

/

100.8

/

April 12, 1978

Infrared Spectra of Argon Matrix-Isolated Alkali Halide Salt/ Water Complexes Bruce S. Ault Contributionf r o m the Department of Chemistry, University of Cincinnati, Cincinnati, Ohio 45221. Received July 6 , 1977

Abstract: The reaction products of alkali halide salt molecules with H20 and its isotopic counterparts at high dilution in argon matrices have been investigated. The infrared spectra of these complexes show two bands in the 0 - H stretching region, and two bands between 400 and 700 cm-l, which show a slight dependence on both the alkali metal cation and the halide anion. These results have been interpreted to indicate that a pyramidal structure is formed in which the metal cation is bound to the oxygen of the water molecule, and that the anion sits as a counterion at the base of the pyramid and interacts through hydrogen

bonds to the water molecule. The two low-frequencyvibrations show a strong hydrogen dependence and are assigned to the two hydrogen deformation modes anticipated for a C, pyramidal structure. The infrared spectra of transition metal aquo complexes, used as a model for these complexes, support the assignment of a metal cation-bound species. Comparison of these results to the theoretical calculations on such complexes is made as well.

Introduction The interactions of alkali metal cations with water and other bases have generated extensive interest over the years, from the biological implications in nerve axons to the solution studies of strong electrolyte^.'-^ Numerous efforts have been made in the past few years to sort out the different aspects of these interactions, and in particular when competitive hydration of both an alkali metal cation and a halide anion can occur. Kebarle and co-workers have thoroughly investigated the gas-phase hydration of both alkali metal cations and halide anions, and determined heats of hydration for one to six water molecule^.^,^ While the heats of hydration for several H20 molecules are more directed toward concentrated aqueous solution, the data for hydration of a single alkali metal cation, or halide anion, are of interest for competitive hydration in an isolated environment, either gas phase or inert matrix. Recently, Devlin and co-workers have looked at the low-temperature spectra of small amounts of water in M'C104glasses, to determine the relative effects of hydrogen bonding to the water and interactions of the metal cations with the water molecules.' These workers have also studied the cation-water interaction through the infrared spectra of alkali nitrate and chlorate ion pairs in glassy H20 matrices. Devlin and co-workers have also investigated argon-matrix-isolated samples of alkali metal nitrates and H20, over the range 6 to 100% H20, focusing their studies on the nitrate anion bands to determine the hydration effects of the water m ~ l e c u l e s . ~ These studies were then related to vibrational spectra of these same alkali metal nitrates in dilute aqueous solution, which have been the subject of numerous investigations over the years. l o Numerous theoretical calculations have been directed toward elucidation of the M+-H20complex, from very simple models to detailed a b initio calculation^.^'-^^ The most sophisticated of the calculations have been able to reproduce well the experimental heats of hydration in the gas phase, as well as the geometry of the complex, and one calculation predicts vibrational frequencies of the complex.'* With the obvious interest in these interactions, from both experimental and theoretical points of view, a study of the matrix isolated complex is suggested. The salt-molecule reaction technique has been used in argon matrices to transfer the halide anion of an alkali halide salt molecule to a Lewis acid, to form an anion of interest in an ion pair.I5-l7 Similarly, reactions of these salt molecules with Lewis bases and, in particular, with H2O might lead to a complex in which the alkali metal cation has transferred to the base to form a 0002-7863/78/1500-2426$01.00/0

M+.H20 complex in an ion pair. Of course, under these circumstances, a hydrogen bonding interaction can also occur, which might represent the competition between the alkali metal cation and the halide anion for the H20 in an isolated ion pair. With all of the theoretical results available, this complex will also provide a comparison of experiment and theory, so a complete study of the interaction of alkali halide salt molecules with H20 was undertaken, with systematic variation of both the alkali metal cation and the halide anion. (Reactions of the alkali fluoride salts with H20 have not been considered in the present work-the unique reactivity of the fluoride anion leads to very different spectra than those presented here and will be reported a t a later date.) Experimental Section All of the experiments conducted for this study were carried out in a completely stainless steel vacuum system, with Nupro Teflon-seat valves. Pumping was provided by a Model 1402B Welch vacuum pump, and a Varian M-2 diffusion pump, with liquid nitrogen trap. Vacuums near lo-' mm at the gauge (cold cathode, Varian) were generally recorded at the start of an experiment. Cryogenics were supplied by a Model 21 closed cycle refrigerator (CTi, Inc.), operating down to 10 K. However, normal temperatures during deposition ranged near 14-16 K. Gas samples were deposited from 2-L stainless steel vessels through a finely metering valve onto the cold surface, a CsI window mounted with indium gaskets to the copper cold block. Deposition of the gas samples was essentially perpendicular to the cold surface. Temperatures at the cold surface were measured with a gold-doped cobalt vs. iron thermocouple, and window temperatures were regulated by supplying a constant voltage to two 10-W button heaters mounted near the cold window. The alkali halide salts were heated and vaporized in a small resistively heated oven employing chrome1 A heating wire, and deposited at approximately a 45' angle to the cold surface.The vacuum vessel was equipped with CsI windows throughout, and sat in the sample beam of a Beckman IR-12 infrared spectrophotometer for the duration of the experiment, to allow monitoring of the sample during deposition. Triply distilled H20 was used for experiments employing normal isotopic H2O. A small amount, 1-2 mL, was put into a glass finger and repetitively frozen, pumped on, thawed, and pumped for several freeze/thaw cycles to remove any volatile species. For deuterated experiments, D20 (99.8%, Merck) was employed with a similar number of freeze/thaw cycles. To guard against isotopic exchange during the experiment, the entire vacuum manifold was conditioned with D2O vapor before preparation of the actual experimental sample. D/H ratios of at least 6 were easily obtained in this manner. In several experiments, '80-enriched H20 was employed (95%, Merck), with the usual freezelthaw cycles. The salts employed were NaCl (Mallinckrodt, analytical reagent), KCl (Allied, reagent), RbCl (Fairmount), CsCl (Fisher), CsBr (Orion), CsI (Harshaw), and KCN 0 1978 American Chemical Society

Ault

/ IR Spectra of Alkali Halide SaltlWater Complexes

I

NaCl

I

I

2427

H-01

I

N

I

RbCl * H20 '

T!

-3200

3i2c

-

2460

-

'3ec

.

L?A\LENL.MBER

Y ,.",

3260

3180

3100

,

710

WAVENUMBER

,

. 670

.

*

*

500

,

. 460

,

,

,

.

1

\ , I _ ,

233c

482

44c

400

360

320

lcm1

Figure 2. Infrared spectra of the reaction products of CsCl with isotopic water molecules in Ar matrices at 14 K, over the spectral regions 300-500, 2260-2480, and 3 180-3240 cm-'

420

(cm-1)

Figure 1. Infrared spectra of each of the alkali chloride salts deposited into an Ar matrix containing HzO at M / R = 600, over the spectral regions 400-500,650-730, and 3060-3280 cm-I, after 20-24 h deposition at 14 K.

(Fisher). Argon and nitrogen were used as matrix gases without further purification. Gas samples, with matrix/reagent ratios ( M / R ) from 500 to 2000, were deposited at a rate of 2 m M / h for 24-36 h before final spectra were recorded. The salts were outgassed in the oven, behind a closed door, for several hours before an experiment, at temperatures exceeding those used during the experiment. The deposition temperatures for these salts were NaCI, 600 OC; KCI, 550 "C; RbC1,525 "C; CsCI, 500 OC; CsBr, 500 OC; CsI, 500 OC; and KCN, 550 "C, although a range of temperatures around these indicated were used to provide a range of concentrations in the matrix. Infrared scans were taken intermittently during the experiment, with final scans after deposition was finished. Final scans consisted of a survey scan from 4000 to 200 cm-', followed by high-resolution scans at expanded scale, of all regions of interest. In many cases, the matrices were then slowly warmed to about 40-44 K to allow diffusion, and then recooled to 16 K and rescanned.

Results Before the spectra of the reaction products of the alkali halide salts with H20 were recorded, the spectra of the HzO or isotopic molecules in Ar alone were recorded carefully. Since H2O has been demonstrated to rotate in an Ar matrix,'8-20 a number of bands were observed in these experiments near the three fundamentals of H20. When a sample of Ar/H20 = 600 was deposited, the spectra revealed bands at 3578, 3715, 3761, and 3781 cm-' in the 0 - H stretching region, and a t 1575, 1595, 1610, 1626, 1639, and 1662 cm-I in the H-0-H bending region. These multiplets are due to rotating H20 in Ar, and are in good agreement with earlier workers. Trace (a) of Figure 4 shows the spectrum of a matrix sample of A r / H 2 0 = 600 over the regions 200-360 and 3560-3800 cm-I. Substitution of DzO yielded the above bands, weakly, of HzO, and bands at 2716 and 3716 cm-I for the 0 - D and 0 - H stretches of HDO, and bands a t 2663,2746,2762,2773, and 2785 cm-l for the 0 - D stretching bands of DzO. Bands were also observed in the bending region a t 1167, 1177, 1183, 1191, and 1197 cm-I for DzO, and a t 1375, 1388, 1392, 1402, 1418, 1432, and 1442 cm-' for H D O . These are all in good agreement with previously reported band positions. When a sample

of Ar/H2l80 = 600 was studied, bands were observed at 3569, 3699, 3748, and 3766 cm-I in the stretching region, and a t 1569, 1588,1603, 1618, 1634, and 1656 cm-lin the bending region. While these bands have not been previously reported in Ar matrices, they all show a few wavenumbers shift to lower energy from their l6O counterparts, and are reasonably assigned to H2180. It is interesting to note, also, that workers have shown that doping an Ar/H20 matrix with another atom or compounds, such as Xe, can partially or entirely eliminate the rotational structure of the H20 vibrational bands, presumably by destroying the local structure of the Ar lattice.2' This effect was observed when NaCl was deposited into these matrices, as well as some of the other salts. In these cases, "nonrotating" H2O bands were observed a t 3720 and 1595 cm-l, in agreement again with earlier workers. Finally, H2O has been shown not to rotate in N2 matrices, and the results here for an N2/H20 = 600 matrix showed bands a t 3554, 3639, 3703, 3720, and 3731 cm-' in the 0 - H stretching region, and a t 1600 and 162 1 cm-' in the H-0-H bending region. These bands are in good agreement with those assigned earlier to the stretching and bending modes of H2O and (H20)2, with intensity ratios suggesting a predominance of the monomer a t these concentrations. CsCl HzO. Two experiments were run in which CsCl was deposited into an Ar matrix without any added H20. However, some residual H20 was present, and judging from the band intensities the A r / H 2 0 ratio was about 3000. In these two experiments, with different CsCl levels, weak bands were observed at 462,697,3116, and 3204 cm-', with the band at 697 cm-I barely perceptible. These bands were somewhat increased in the experiment a t higher CsCl level, and appeared to maintain a constant intensity ratio. The two following experiments involved CsCl Ar/H20, with M/R = 1000 and 600. In both experiments, the four bands listed above grew considerably in intensity, with optical densities of 0.21,0.06, 0.44, and 0.10, respectively, for the 462,697, 31 16, and 3204 cm-I bands, at M/R = 1000. All four of these bands were also quite sharp, as can be seen in traces (a) and (b) of Figure 2. In addition, weak, broad bands were observed a t 542,3175, and 3310 cm-l, with OD values less than 0.05. At an M / R of 600, the four bands grew further, and maintained a constant intensity ratio, while the second set of bands a t a t 542,3175, and 3310 cm-' grew more rapidly. When this sample was annealed

+

+

Journal of the American Chemical Society / IOO:8

2428

/ April

12, I978

temperature, in that a lower salt level in the matrix is produced, leading to lower yield of reaction product. KCI H20. KC1 was deposited into an Ar matrix in one experiment, with a small amount of impurity H2O present, and a band at 248 cm-l was observed which has been assigned previously to Ar matrix-isolated KC1 monomer.22 In addition, weak bands were observed a t 475, 3117 and 3201 cm-I, analogous to the bands in the CsCl/Ar experiment. When an experiment was carried out, reacting KCI and Ar/H20 = 600, these three bands grew in intensity, and two new weak bands were observed, at 688 and a t 242 cm-I, as a distinct shoulder --3210 3 i.F C , . . -“v , __ Ld 2420 23-.3 225:, L@C. 44; 4CO 36i 32C on the parent KCI monomer. Also, weak bands were observed YIAIELIJCIBEQ icm’: a t 410, 546, 3229, 3288, and 3338 cm-l. An experiment was Figure 3. Infrared spectra of the reaction products of CsBr with isotopic also conducted with H2180, but the overall yield was low in this water molecules in Ar matrices at 14 K , over the same spectral region as experiment. Product bands were observed a t 473, 3107, and Figure 2. 3188 crn-l, but no I8Ocounterparts were detected for the 688and 242-cm-’ bands. N o deuterium substitution was atto 42 K, the second set of bands grew slightly, while the original tempted with KCI. four bands decreased a small amount. NaCl H20. Six experiments were conducted in which Substitution of I80-enriched H20 in one experiment a t NaCl was deposited into an Ar matrix without any added H2O. M / R = 600 led to a shift in each band position, to 460, 695, Some H 2 0 , nonetheless, was present with the M / R around 3 108, and 3 194 cm-l, while the second set of bands (the bands 3000. In these experiments, varying the oven temperature for at 542,3175,3254, and 3310 cm-I in the I6Oexperiment) was the NaCl evaporation, three major bands were observed, a t relatively weak and shifted to 541,3164, 3242, and 3301 cm-’. 469, 3141, and 3251 cm-I, in addition to parent NaCl and The l6O counterparts were also weakly observed owing to the (NaC1)z bands at 333,275, and 226 cm-l. These bands showed imperfect enrichment of the H2l80 sample, and impurity H20 constant relative intensity through the differing experiments, in the system. These bands are displayed in trace (c) of Figure including several experiments in which the matrices were de2. posited a t different cold window temperatures, ranging from Deuterium substitution in several different ratios was used 10 to 21 K. This temperature variation appeared to have very in an attempt to clarify the nature of the reaction products. In little overall effect on product formation. Also, two weak bands the first experiment, with an approximate D / H ratio of 6 were observed at 258 and 290 cm-I. These have been reported (taken from the parent band intensities), a distinct counterpart by previous workers,22 and assigned to aggregate species of was observed for each of the original four bands listed above, NaCI, although such a characterization would be difficult on in addition to the parent bands of D20, HDO, and H2O. These the basis of the data obtained here. In addition, weak broad new bands were observed at 342,497, 2318, and 2382 cm-l. bands were detected in the ranges 400-440, 510-520, and These bands retained, even on deuterium substitution, ap3320-3450 cm-l. When NaCl was then deposited into maproximately the same intensity ratios. In addition, a weak band trices of A r / H 2 0 = 600, the three major bands grew considwas observed at 2340 cm-I, and the bands observed in the pure erably, and maintained a constant intensity ratio throughout, H experiment were observed very weakly from H20 impurity and the weak broad bands grew as well, but at a different rate in the system. An experiment was then run in which the D/H than the three major bands. The infrared spectra of all four of ratio was adjusted to approximately 4. In this experiment, the the alkali chloride salts with H2O can be seen in Figure I . bands observed in the nearly pure deuterium experiment reTrace (b) of Figure 4 shows the spectrum of a sample of NaCl mained, with similar intensity; the bands due to the pure H A r / H 2 0 , in the NaCl stretching region and in the uncomspecies grew slightly; and bands were observed a t 2340, 3142, plexed H20 region, 3560-3800 cm-’. and 401 cm-I. The band at 2340 cm-’, which had been much NaCl was also deposited into an N2 matrix in one experiless intense than its neighbors a t 2318 and 2382 cm-I in the ment, and bands were observed a t 310, 259, and 220 cm-], which must be the bands analogous to the three NaCl and experiment with D / H = 6, was now equal in intensity to the (NaCl)2 parent bonds observed in Ar. When NaCl was de2382-cm-I band in this D / H = 4 experiment. This experiment posited into a matrix of N z / H 2 0 = 750, three new bands were is shown in trace (e) of Figure 2. observed, at 465, 3 160, and 3265 cm-l. These bands were less A final experiment was then conducted in which a D / H ratio intense and somewhat broader than the analogous product of about 1 was used. In this experiment all of the pure D bands bands in Ar matrices. When this sample was annealed to 37 were observed, all of the bands observed in the pure H experK and recooled to 16 K, all three bands grew slightly and in iment earlier were also observed, and the new bands at 401, same relative ratios. Since the overall yields in N2 were low, 2340, and 3 142 cm-’ grew considerably. Three sets of bands no further experiments were conducted in N2 matrices. then were observed, one of which was observed in the pure H20 CsBr. Several experiments were carried out in which CsBr experiments, one of which predominated a t very high D2O was deposited with samples of A r / H 2 0 and various isotopic levels, and a third set (401, 2340, and 3142 cm-’) which was variations. Three major product bands were observed in the most evident a t intermediate D / H levels. These three sets reaction of CsBr with a sample of A r / H 2 0 = 600, a t 435, seemed to be internally consistent with respect to intensities 3 158, and 3229 cm-I. These bands were all relatively intense over the entire range of experiments. and sharp, and resembled closely the product bands in the CsCl RbCl H20. RbCl was deposited into an argon matrix H 2 0 experiments, with slight shifts in band position. Weak, containing H2O with M / R = 600 in two different experiments. broad aggregate bands were also observed in the regions In addition to the RbCl monomer parent band a t 206 cm-I 3240-3330 and 390-420 cm-I. Several deuteration experithree new product bands were observed, at 47 1 , 31 10, and 3202 ments were carried out employing CsBr and samples of Ar/ cm-I. In both experiments, the bands were weak but reproD20, with varying degrees of deuteration. In an experiment ducible, and in general agreement with bands observed in the with D / H = 4, the 435-cm-’ band shifted to 321 cm-], and CsCl HzO reaction. The overall low yield was probably due a weaker band appeared a t 373 cm-’, while in the high-freto poor oven alignment, rather than overall low reactivity. The quency region, two new bands appeared a t 2338 and 2395 effect of oven alignment is similar to that of lowering oven

+

,av1

1

-

---A-

+

+

+

+

+

Ault

2429

IR Spectra of Alkali Halide SaltlWater Complexes

Table I. Band Positions (cm-I) of Salt Molecule/HzO Reaction Products in Argon Matrices

NaCl KCI KCI KCN RbCl

258 242?

CsCl CSCI"

CsClb CsBr CSI a

D2O.

H2180.

475 473 512 47 1 462 342 460 435 40 1

688

697 497 695

3141 3117 3107 3040 3110 3116 2318 3108 3158 3164

3251 3201 3188 3165 3202 3204 2382 3194 3229 3280

Possible assignment.

cm-', the latter appearing to have a shoulder on the low-energy side. I n addition, a weak band was observed near 3198 cm-I, in addition to the weak bands a t 3158 and 3229 cm-' due to the pure H species. Lowering the deuterium ratio to D / H = 1 caused the three low-frequency bands a t 435,373, and 321 cm-' to appear as a distinct triplet, approximately 1:2:1. In the high-energy region, the bands at 2338 and 2395 cm-' were weakened, the shoulder a t 2391 cm-I grew, and resolved into a distinct band, and the weak band a t 3198 cm-I grew somewhat, while the 3158- and 3229-cm-' bands grew considerably. These results are all presented in the three traces of Figure 3. CsI H20. CsI was vaporized and deposited into an Ar matrix in one experiment, and into a matrix containing Ar/ H2O = 600 in a second experiment. No infrared fundamental of the parent monomeric CsI molecule was expected owing to its low frequency, and this was the case. However, three weak bands were observed in the former experiment, a t 407,3164, and 3280 cm-I, and these bands were again observed in the latter experiment with considerably greater intensity. No deuteration experiments were attempted with CsI. KCN H2O. KCN was used in several experiments to determine the role of the halide anion in the complex. KCN has been investigated previously in argon matrices by Margrave and co-workers, and the spectra obtained here for the deposition of KCN into a blank Ar matrix were in good agreement with their results for both monomer and dimer bands.22The matrix was a characteristically deep blue color in all experiments employing KCN. When KCN was codeposited with a sample of A r / H 2 0 = 600, new bands grew in, 3040,3165, and a doublet a t 502 and 512 cm-l, with intensity ratios approximately the same as the analogous bands in the KCI Ar/H20 experiments.

+

+

+

Discussion The results from the different salt/HzO experiments can be summarized by observing that strong bands were formed upon reaction of salt molecules with H 2 0 , and that from the bands obtained two distinct groups may be identified. The first group, typified by the bands at 462,697,3116, and 3204 cm-I in the CsCl H2O experiments, appeared even in the most dilute experiments, when water was present only as an impurity, and grew considerably when H20 was added in subsequent experiments. These four bands also showed the same relative intensities throughout all of the experiments, and the behavior of this group of bands upon dilution indicates that they can be assigned to a 1 : 1 reaction product of the salt molecule with H2O. The second group of bands, typified by the bands a t 542,3175,3254, and 3310 cm-' in the CsCl H2O experiments, appeared only a t higher concentrations of the reactants, and particularly a t higher concentrations of H20. This behavior, coupled with the fact that these bands grew during annealing, suggests that they may be assigned to an aggregate species rather than a well-defined 1:l reaction

+

+

I

U

h

3720

3640

360

WAVENUMBER

(cm-1)

320

280

240

Figure 4. Infrared spectra of matrix samples over the spectral ranges 200-360 and 3560-3800 cm-I. Trace (a) shows a sample of Ar/H*O = 600, while trace (b) shows a sample of Ar/H2O = 600 NaCI. Trace (c) depicts a sample of CsBr A r / H 2 0 = 600, over the region 3560-3800 cm-I.

+

+

product. This assignment is also suggested by the broadness of the infrared absorption bands. This set of bands had a counterpart set at 541, 3164, 3242, and 3301 cm-' in the CsCl Ar/H2I80 experiment, and these bands must be assigned to the I8O analogue of the aggregate species. Not much can be surmised about the nature of the aggregate species, and the following discussion will focus on the three or four bands in each experiment which are associated with the 1:l reaction product, and whose positions are listed in Table I. In each experiment, with different alkali halide salts and H20, two bands were observed in the region 3100-3300 cm-I, while one or two bands were observed between 400 and 700 cm-I and, in the case of KCI HzO, a weak band was observed below 300 cm-'. The region above 3100 cm-l can reasonably be characterized as the 0-H stretching region and, for a 1:l complex, two such bands are expected, originating from the two 0 - H stretching modes of the H20 precursor molecule. In the case of CsCl H2O these bands are located at 31 16 and 3204 cm-I, while Table I lists the band positions for the remaining alkali halide/H20 reaction products. The deuterium results support this assignment-with a distinct deuterium counterpart observed for each band. The 3 116-cm-I band apparently shifts to 2318 cm-I, and the 3204-cm-' band shifts to 2382 cm-'. These shift ratios, 1.34 and 1.34, respectively, are of a magnitude expected for an 0 - H stretching vibration, and are very close to the shift ratios for v1 and v 3 of H2O. Also, the 2382-cm-l band has the same intensity relative to the 2318-cm-' band that the 3204-cm-I band has to the 3 1 16-cm-' band, supporting this mapping as well. Equally important, as the D / H ratio was reduced, and as more hydrogen was introduced into the system, two new bands grew in, one between the two deuterium counterparts a t 2345 cm-I, and the other between the two hydrogen counterparts at 3142 cm-l. These bands have the right position and intensities to be assigned to the two stretching modes of CsCl/HDO complex. In such a complex one would expect one vibration of nearly pure 0 - H character and one of nearly 0 - D character. This is exactly the behavior observed here and is precisely the behavior of isolated HDO, where one stretching mode is near that of D 2 0 and the other is near H20. The above results suggest strongly that the reaction complex contains two equivalent hydrogen atoms and, if this were not the case, then a considerably different spectrum would have been obtained. It is apparent from the spectra that there is an intensity reversal of v~ and u 3 from the spectra of the free H2O molecule. v 3 is the more intense band for H20, yet the band observed here for the complex which correlates with v1 is considerably more intense. This effect has been observed previously and discussed

+

+

+

Journal of the American Chemical Society

2430

/

100:s

/

April 12, 1978

Table 11. Gas-Phase Energetics of Alkali Metal Cation and Halide

Anion Hydrationo M+ + H20

M+

(H2OM)+ - A H . kcalimol

Li+

34.0 24.0 17.9 15.9 13.7

Na+

Kf Rb+ CSf

X-

X-

X-

+ HzO

+

(XH20)-

- A H . kcal /mol

FClBr-

23.3 13.1 12.6 ~n

T-

I. Dzidic and P. Kebarle, J . Phys. Chem., 74, 1466 (1 970); M. Arshadi, R. Yamdagni, and P. Kabarle, ibid.,74, 1475 (1970). Figure 5. Four possible structures for the 1:1 complex of alkali halide salt molecules with H2O isolated in argon matrices.

in terms of the Ferguson-Matson-Friedrich-Person model,23 which demonstrates that an intensity enhancement is expected for totally symmetric vibrational modes of charge transfer complexes. As a consequence, this result is not unexpected here, and points toward a charge transfer or ion transfer complex. The results employing I8O-enriched H2O suggest also that the reaction occurring is forming a 1:l complex. The shift in the band positions with H2I80 is almost the exact same shift as observed for free H2180 relative to H2I60, again suggesting an 0 - H stretching mode. More importantly, a small amount of residual H2I60 was present in these experiments and, in the case of the 3204-cm-' band, a distinct doublet was observed with no intermediate bands suggesting a species with only a single water molecule. It is, of course, possible that the complex contains two water molecules, whose vibrations are nearly completely uncoupled, but the concentration studies, along with the I8O data, tend to refute this possibility. One must also consider the possibility that the observed complex is between the salt dimer, (MX)2, and a water molecule, since the salt vapor over the solid at this temperature does contain a considerable fraction of the dimer. This is not likely, however, since the dimer is very strongly bound, and the strength of interaction with H2O is not great enough to disrupt the dimer structure. Also, if the salt dimer/HlO complex accounted for the observed product species, it would imply that a complex is not formed between the salt monomer and H 2 0 , which is not a likely circumstance. In experiments with alkali halide salt molecules and H 2 0 , one or two bands were also observed in the region 400-700 cm-I. Specifically, a band was observed between 400 and 480 cm-' in each case, and in several experiments a second band was observed near 700 cm-I. The 1 : 1 reaction complex, almost independent of geometry (which will be discussed in a later section of the paper), will be expected to have two deformation modes which, owing to the lightness of the hydrogen atoms, should be essentially hydrogen vibrations. The deuterium shifts in the CsCl experiments suggest that this is the case, with a strong deuterium shift observed for both the 462- and 697cm-l bands, to 342 and 497 cm-l. The shifts here are 1.35 and 1.40, respectively. Again, the mixed D / H experiments are very informative, in that a distinct triplet is observed for the 462cm-l band. The pure deuterium counterpart was observed a t 342 cm-' and an intermediate band grew in with decreasing D / H ratio, at 401 cm-l. This distinct triplet lends compelling support for the conclusion that the 1:1 complex contains two equivalent hydrogen atoms. This should be the case as well for the 697-cm-I band and its deuterium counterparts, but this

was the weakest band system detected and no intermediate band was detected, presumably owing to low intensities. The final band observed in this sequence of experiments was a weak band appearing as a shoulder in the low-energy side of the 248-cm-' KC1 parent band on the KC1 H2O experiment in Ar. This band must be associated with a product species as well, as it is not detected in the KCl Ar experiments, but is quite clear in the KC1 A r / H 2 0 experiments. Depending on the nature of the complex, this band may either be interpreted as a perturbed KCl vibration in the complex or as, perhaps, a K - 0 stretching vibration. Consideration of the structure of the complex, in the next portion of the paper, may aid in the assignment of this band. Regardless of this assignment, the bands associated with the 1:l salt/H20 complex can be assigned to two 0 - H stretching vibrations, a hydrogen deformation vibration (or two in the case of KC1 and CsCl H20), and possibly a metal-oxygen stretching band.The evidence from all of these band assignments also clearly supports a complex in which the two hydrogen atoms are equivalent. Determination of the structure or geometry of the complex is of considerable interest as well. The gas-phase heats of reaction from Kebarle's work are tabulated in Table 11, and these values suggest that the binding of an alkali metal cation to the lone pairs on H20 is generally stronger than the hydrogen bonding of the halide anion to the hydrogens on the water molecule, with the exception of the fluoride anion, which has not been considered here. The use of ion cyclotron resonance (ICR) has allowed for further determination of the strengths of binding of alkali metal cations to a large number of bases,24 suggesting that binding energies up to 50 kcal/mol can occur for Li+. Theoretical calculations also tend to support the binding of a cation preferentially over the hydrogen bonding of the anion. Four possible structures can be envisioned for such a complex, and are depicted in Figure 5. These are undoubtedly not all of the possibilities but are the major ones. Structure 4, which employs an unsymmetrical hydrogen bond, can be ruled out quickly on the basis of equivalence of the hydrogen atoms, which has been demonstrated above. If this were, in fact, the geometry, one would expect to see a single hydrogen stretching mode of the hydrogen-bonded hydrogen shifted well down from H20 and another hydrogen stretching frequency nearly equal to that of free H20. A careful search for such a band was made, and no such band was observed. Figure 4 shows this search for the NaCl A r / H 2 0 and CsBr A r / H 2 0 experiments by comparison to trace (a), A r / H 2 0 with no added salt. This is in agreement with the fact that two bands were observed considerably shifted from free H20.

+

+

+

+

+

+

/ IR Spectra of Alkali Halide SaltlWater Complexes

243 1

Structure 3, involving a symmetrically hydrogen-bonded species, fulfills the requirement of two equivalent hydrogen atoms. The vibrational modes of this complex should show virtually no dependence on the nature of the metal cation, as the cation simply serves as a counterion for stability. However, a dependence was observed, although relatively small, as shown in Table I. Secondly, the 0-H stretching modes should show a strong dependence on the halide anion used in the complex, since the hydrogen bonding capabilities of the halide anions vary considerably from C1- to I-, and thus the band positions should shift accordingly. According to the relation proposed by Pimentel and c o - w o r k e r ~this , ~ ~band position should depend directly on the proton affinity of the base or anion, and these are known to range from 333 kcal/mol for C1- to 313 cm-' for I-.26 The results with K C N show this as well, since CNshould have considerably different hydrogen bonding characteristics, yet KCNaH20 showed a similar spectrum to KCl H20. These band positions are also listed in Table I. One can make an estimate of the 0 - H stretching frequencies in a purely hydrogen-bonded complex such as structure 3 by comparison of the hydrogen stretching frequencies of a CI- anion hydrogen bonded to NH3 (a weaker acid) and to HCI, a stronger acid, since the shift in the hydrogen bond vibrational frequency of a series of acids with a single Lewis base has been shown to depend monotonically on the acid strength of the hydrogen donor.25 In the former case, the N-H-C1stretching band27,28fell between 1900 and 2600 ern-', while in the latter case,I6 the C1-H-C1- band fell near 700 cm-'. H 2 0 , with intermediate acidity and hydrogen bonding character, should yield a band between these two extremes, if structure 3 is the correct geometry. This is not observed, as the bands were detected in the 3 100-cm-I region suggesting that a purely hydrogen-bonded species is not formed. One theoretical calculation has been performed on the Cl-sH20 system, and indicated that the bonding was largely electrostatic, with little charge migration.29However, this work did not predict frequency shifts for the hydrogen stretching frequencies, and cannot be used for direct comparison here. Generally, the band shapes of hydrogen-bonded stretching vibrational bands are quite broad,30 for both solution phase hydrogen-bonded species and matrix-isolated hydrogenbonded acid/base pairs.16,31s32,33 This is not the case heie, suggesting again that the interaction is not purely a hydrogen bonding interaction. Gas-phase energetics argue against structure 3 since the interaction of an isolated metal cation with H20 has been shown to be stronger than an isolated halide anion with H20. All of these points together serve to rule against the symmetrically hydrogen-bonded complex. The first complex, 1, in which a planar cationic species is formed through the interaction of the metal cation with the two lone pairs of the oxygen, satisfied the requirement of two equivalent hydrogens as well. However, the position of the 0 - H stretching bands, near 3100-3200 cm-', rules against this structure. It would be hard to envision a complex in which the alkali metal cation withdraws sufficient elebtron density from the 0-H bonds to lower the vibrational frequency about 600 cm-I. Recent studies have been directed a t determining the effect of an isolated cation on the vibrations of the water molecule, and the best estimates available now suggest that the 0 - H stretching frequencies will shift a few wavenumbers a t most.34The magnitude of these shifts are more indicative of some hydrogen-bonding interaction, where very large shifts are common. Moreover, the band positions do show a slight halide anion shift, which would not be expected for a planar cationic species where the halide anion is serving strictly as a counterion. This planar structure then is not probable either. The final structure depicted here, 2, satisfies all of the points mentioned above. In this complex, the alkali metal cation binds

to the oxygen of the water molecule, as suggested by gas-phase energetics. However, the cation formed is pyramidal rather than planar, brought about by the interaction with a single lone pair on the oxygen, rather than by equal interaction with both lone pairs. The addition of an H+ to H2O to form H 3 0 +yields a pyramidal species, so it is not surprising that the addition of a metal cation M+ to H20 should lead to a pyramidal species. I n this pyramidal configuration the halide anion serves as a counterion, but in doing so can interact with all of the centers of positive charge in the complex, including the hydrogens as well as the alkali metal. This would lead to a partial (nonplanar) hydrogen-bonding interaction, and shifts of the 0 - H stretching frequency comparable to those observed. This structure should also lead to both a dependence on the cation and on the anion for all of the vibrational modes, as was the case. The gas-phase energetics also would support this particular structure, and the observation of a metal-oxygen stretching frequency could be accounted for as well. This geometry does not require the anion to sit symmetrically at the base of the pyramidal structure, but rather it should favor the center of largest positive charge, which will undoubtedly still be the metal atom. The geometry will be C, in this complex, and this particular structure fits all of the available data. It is also worth noting that in the gas-phase water dimer, (H20)2, the structure is open, and the oxygen atom involved in the hydrogen bond is bonded in a pyramidal arrangement, not in a planar c o n f i g u r a t i ~ nThis, . ~ ~ too, supports the structure proposed here. One other possible structure is a planar structure in which the metal cation binds to the oxygen, while the halide anion symmetrically hydrogen bonds to the two hydrogens of the water molecule, a composite of structures 1 and 3. However, this structure can be ruled out by observing that it is very unlikely that the major centers of positive and negative charge would be so completely separated, when the pyramidal geometry would allow for considerably greater attractions. Moreover, it is hard to envision how such a complex could be formed, for it would require the complete rupture of the M-CI bond during complex formation. The pyramidal C, structure for the complex should have, as noted earlier, two hydrogen stretching modes, which are observed. The complex should also have two hydrogen deformation modes, originating with the A I mode of the pyramidal C3c molecule, and one of the components of the doubly degenerate E mode of the C3Lmolecule, since the degeneracy is broken upon reduction of the symmetry to C,. 36 This accounts for the two deformation modes observed for a number of the complexes. For many of the complexes only one deformation mode was observed, while the second was probably not observed for lack of intensity. When this second band was observed, as in the CsCl t H20 experiment, it was quite weak. In addition, this pyramidal species should show a metal-oxygen stretching mode, but this would likely lie below 200 cm-l, the instrumental cutoff, except possibly for the N a + and K+ complexes. Such a band may have been observed in the KC1 experiment at 242 cm-I, while in the NaCl experiment, the only candidate is the band at 258 cm-I. This has been assigned by previous workers to an aggregate species of NaC1.22However, if these workers also had H20 present as an impurity, then the band may have been misassigned. Insufficient evidence is available to make a definitive assignment of this band, either to an aggregate species or to the N a - 0 stretch of the NaClmH20 complex. The infrared spectrum of crystalline Li2S04-H20 has been investigated, and the complex is shown to be bound through Li+ cation to the water molecule.37In the spectrum, two librational modes of the water molecule were observed between 400 and 600 cm-], analogous to the two deformation modes observed here. These workers also observed the L i - 0 stretching band near 400 cm-l, which leads to a

Ault

+

2432

Journal of the American Chemical Society

/

100:8

/

April 12, 1978

prediction, on the basis of mass and force constants, of The third study was done by co-condensing alkali nitrate frequencies below 200 cm-' for the metal-oxygen stretching salts in argon matrices doped with H 2 0 , in some respects frequency in most of the complexes studied here. Finally, one similar to the studies here.9 They concluded that the cation would expect a hydrogen bending mode analogous to the bound to the H20 in all cases, and that the salt M+NO3- re1600-cm-I bending mode of free H 2 0 , perhaps in the region mained a contact ion pair, in qualitative agreement with the near 1600 cm-'. Such a band was not observed, although a results here. It should be noted, though, that these studies were careful search was made. The lack of observation could be due carried out a t much higher H20 levels, 6 to 100%HzO, while either to lack of intensity or lack of a shift from free H 2 0 , the work here, aimed a t the 1:l complex, was carried out a t which contributed a number of bands in the 1600-cm-l region. 0.10-0.02% H20 in Ar. Also, Devlin and co-workers deposited Regardless, lack of observation of such a band cannot detract for much shorter times, and monitored the very intense N 0 3 significantly from the overall assignments made here. bands as a measure of the interaction. Consequently, they did One cannot overlook the possibility that more than one not observe bands in the 3100-cm-' region analogous to those complex is formed, but owing to band shifts and/or absorption observed in this work. It may be presumed, though, that under coefficients, only one form is detected. This seems unlikely, similar conditions of deposition, concentration, and length, however, since even a 2-4-cm-l shift would have been obsimilar bands would have been observed. served, since all bandwidths were quite sharp. A careful search A comparison to the numerous theoretical efforts concerning was carried out in the region of uncomplexed water particuthis interaction should be made as well.] ] - I 4 These calculations larly, as evidenced in Figure 4, and no new bands were obwere generally made for Li+ (although N a + and K+ were served, even with very small shifts from H2O. The possibility considered in one case),12 and the geometry of lowest total of a complex with very low absorption coefficients remains, but energy was found to be planar. However, this does not take into in view of the ionic nature of the complex, lack of observation account the presence of a halide anion as a counterion, so that of a major product species for this reason is not likely. the comparison in geometry is not entirely valid. Clernentil2 Transition metal aquo complexes, of the form M ( H Z O ) ~ ~ + , also predicts vibrational frequencies for the metal-oxygen can serve as a model for predicting vibrational frequencies in stretching vibration, and the two deformation modes. H e alkali metal/H20 complexes. In the transition metal compredicts the two deformation frequencies in the 300-400-cm-' plexes, two hydrogen deformation modes have been obrange, with slight dependence on the metal cation, which is in ~ e r v e d , ~ ~one - ~between O 450 and 650 cm-' and the other bereasonable agreement with the observed frequencies, 462 and tween 650 and 850 cm-'. Since the interaction between an 697 cm-' for the CsCI/H20 complex. H e also predicts that alkali metal cation and H2O is presumably slightly weaker the metal-oxygen stretching frequencies will lie quite low, below 250 cm-I, except for Li+, which was not studied here. than the interaction between a transition metal cation and H20, one might predict slightly lower frequencies for the Clementi's Mulliken population analysis suggests that very M + H 2 0 complexes postulated here, and this fits well with the little electron density is transferred to the metal cation (about one or two low-frequency bands observed in this work. Also, 0.01 e-), and some electron density is transferred to the oxygen from the hydrogen atoms (0.1-0.2 e-), leaving the metal cation the metal-oxygen stretching mode has been observed for the still the largest center of positive charge.I2 Also, this predicts transition metal complexes, and the band falls between 300 and withdrawal of electron density from the hydrogen atoms from 500 cm-' in each case. With the slightly weaker interaction, which one may infer a weakening of the 0 - H bonds and a and subsequently lower force constant, values below 300 cm-I subsequent lowering of the vibrational frequencies as observed, are not unexpected for N a + and K+, and below 200 cm-l for although the major portion of the lowering of the 0 - H the remaining alkali metal cations. The model of the transition stretching frequency must come through hydrogen bonding. metal aquo complexes, which are known to be bound through the oxygen lone pairs to the transition metal, fits the observed spectra reasonably well, and supports assignment to a complex Conclusions which does involve metal cation-H20 interactions, such as In each of the reactions of salt molecules with HzO in Ar structure 2 . matrices, three or four major bands were observed which could The results of Devlin and co-workers should be considered a t this point. In three separate studies they have investigated be assigned to 1:1 complex, while additional bands were obthe infrared spectrum of salt-water complexes. In the first served a t higher concentrations of reactants, and have been assigned to aggregate species. The observed bands of the 1: 1 study,' H2O was deposited into glasses formed by the evapocomplex were assigned to the two 0-H stretching modes of the ration and condensation of alkali perchlorate salts. These complex, and the one or two low-frequency bands to the hymixtures, at low temperatures, yielded two band systems in the drogen deformation modes in the complex. The favored ge0 - H stretching region. One set was relatively sensitive to the ometry of the complex on the basis of all of the evidence nature of the alkali metal cation, while the other was not. This available is a pyramidal geometry in which the metal cation was interpreted to indicate that two types of interactions were is bonded to the oxygen atom of the water molecule and the occurring-an inner sphere coordination of the metal cation halide anion sits as a counterion below the base of the pyramid, to the water and an outer sphere hydrogen bonding interaction of the H 2 0 to the anion. This supports the observation here that interacting with the metal cation and the hydrogens of the water molecule. This geometry predicts all of the observed both types of interactions can occur, and that the metal-base features of the infrared spectrum of these complexes, and is interaction is quite strong. It must be noted, however, that this in qualitative agreement with the theoretical calculations on set of experiments by Devlin and co-workers is quite different in nature from those conducted here. alkali metal cation-H20 complexes. In the second study,* Devlin and co-workers deposited alkali Acknowledgments. The author gratefully acknowledges metal nitrates and chlorates into glassy H2O a t 12 K, and support of this research through Grant 8305 from the Research measured the strength of the interaction through the splitting Corporation and the University of Cincinnati Research of the u3 band of N03- and the decreasing frequency of u1 of Council. C103-. They conclude that the polarizing power of the cation is sharply reduced via interaction with the H20 molecules, and References and Notes that the interaction is one in which the cation directly interacts (1) C. U. Smith, "The Brain, Towards an Understanding", Putnam, New York. with the lone electron pairs on the water molecule, as is posN.Y., 1970. (2) "Alkali Metal Complexes", Struct. Bonding (Berlin), 16 (1973). tulated for structure 2.

Grant, Streitwieser

/

Reaction of Uranocenes with Nitro Compounds

(3)K. Balasubrahmanyam and G. J. Janz, J. Solution Chem., 1, 445 (1972). (4)D. E. Irish, lonic Interact., 2, 187 (1971). (5) I. Dzidic and P. Kebarle, J. Phys. Chern., 74, 1466 (1970). (6)M. Arshadi, R . Yamdagni. and P. Kebarle, J. Phys. Chern., 74, 1475 (1970). (7)G.Ritzhaupt and J. P. Devlin, lnorg. Chem., 16,486 (1977). (8)(a) N. Smyrl and J. P. Devlin, J. Phys. Chem., 77,3067(1973);(b) J. Chem. Phys., 81, 1596 (1974). (9)G. Ritzhaupt and J. P. Deviin, J. Phys. Chem., 79,2265 (1975). (IO)See, for example, D. E. Irish and A. R. Davis, Can. J. Chem., 46,943 (1968); J. Braunstein, J. Phys. Chem., 71,3402 (1967). (11) E. Clementi and H. Popkie, J. Chem. Phys., 57, 1077 (1972). (12)H. Kistenmacker, H. Popkie, and E. Clementi. J. Chem. Phys., 58, 1689 (1973). (13)K. G. Spears and S. H. Kim, J. Phys. Chem., 80, 673 (1976). (14)K. G. Spears, J. Phys. Chem., 81, 186 (1977). (15) B. S.Ault and L. Andrews, J. Am. Chern. SOC., 97,3824 (1975). (16)E. S.Ault and L. Andrews, J. Chern. Phys., 63,2466 (1975). (17)B. S.Aul!and L. Andrews. J. Am. Chem. SOC.,98, 1591 (1976). (18)E. Catalan0 and D. E. Milligan, J. Chem. Phys., 30, 45 (1959). (19)J. A. Glaser, J. Chem. Phys., 33,252 (1960). (20)R . L. Reddington and D. E. Milligan, J. Chem. Phys., 37, 2162 (1962). (21)G. P. Ayers and A. D. E. Pullin, Spectrochim. Acta, Part A, 32, 1629,1641 (1976). (22)2. K. Ismail, R. H. Hauge, and J. L. Margrave, J. Mol. Spectrosc., 54,402 (1975).

2433

(23)H. 8. Friedrich and W. 8.Person, J. Chem. Phys., 44,2161 (1966). (24)R. H. Staley and J. L. Beauchamp, J. Am. Chern. SOC., 97, 5920 (1975). (25)B. S.Ault, E. A. Steinback, and C. C. Pimentel, J. Phys. Chem., 79,615 (1975). (26)J. L. Beauchamp, Annu. Rev. Phys. Chem., 22,257 (1971). (27)M. P. Marzocchi, C. W. Fryer, and M. Bambagiotti, Spectrochirn. Acta, 21, 155 (1964). (28)J. A. Frietag, M.S. Thesis, University of Cincinnati, 1966. (29)H. Kistenmacher, H. Popkie, and E. Clementi, J. Chem. Phys., 58, 5627 (1973). (30)G. C. Pimentel and A. L. McClellan "The Hydrogen Bond", W. H. Freeman, San Francisco, Calif., 1960. (31)B. S. Ault and G. C. Pimentel, J. Phys. Chem., 77,57 (1973). (32)B. S.Auk and G. C. Pimentel, J. Phys. Chem., 77, 1649 (1973). (33)P. T. T. Wong and E. Whalley, J. Chern. Phys., 82,2418 (1975). (34)H. P. Hayward, private communication. (35)J. A. Odutola and T. R. Dyke, 32nd Symposlum on Molecular Spectroscopy. Columbus, Ohio, 1977,Paper RC6. (36)K. Nakatwto, "Infrared Spectra of Inorganic and Coordination Compounds", Wiley-interscience, New York, N.Y., 1970,p 95. (37)S.Meshitsuke. H. Takahashi, and K. Higasi, Bull, Chem. Sot. Jpn., 44,3255 (1971). (38)I. Gamo, Bull. Chem. SOC., Jpn., 34,760,765,1430,1433 (1961). (39)G. Sartori, C. Furlani, and A. Damiani, J. lnorg. Nucl. Chem., 8, 119 (1958). (40)I. Nakagawa and T. Shimanouchi, Spectrochim. Acta, 20,429 (1964).

Reaction of Uranocenes with Nitro Compounds1 Charles B. Grant and Andrew Streitwieser, Jr.* Contributionfrom the Department of Chemistry, Unioersity of California, Berkeley, California 94720. Receiaed August 20, I977

Abstract: Uranocenes (di-$-cyclooctatetraeneuranium) are relatively stable to many neutral oxygen-containing organic compounds but react rapidly with aromatic and aliphatic nitro compounds to liberate the cyclooctatetraene ligand in quantitative yield and form azo compounds, often in good yield but in some cases with formation also of the corresponding amines. p-Nitrotoluene reacts more slowly than nitrobenzene, Additional studies of reaction mechanism show that free nitro radical anions or nitrenes do not appear to be involved, but free nitroso compounds are probable intermediates. Azoxy compounds react more slowly with uranocenes and cannot be intermediates in the reactions of nitro compounds. The reaction has few analogies.

Introduction Since the discovery of uranocene in 1968,2 we and others have investigated the generality of the synthesis as well as the physical and chemical properties of the parent compound and its derivative^.^-^ Studies have been made of hydrolytic reactions of u r a n o c e n e ~and , ~ ~some ~ ~ studies of alkyllithium substitution on the carbon ring of appropriately substituted uranocenes have been r e p ~ r t e d .This ~ ~ . paper ~ extends uranocene chemistry into synthetic organic chemistry, focusing on its properties as a reagent for the reduction of nitro compounds and related derivatives. In addition to synthetic potential we have also studied corresponding reaction mechanisms because organoactinide chemistry is a relatively new field and few such studies are presently available. Results In screening tests, acetone, dimethyl sulfoxide, phenyl isocyanate, iodobenzene, styrene oxide, hexamethylphosphoric triamide, ethylene carbonate, and trimethyl phosphate were treated with 1,l'-di-n-butyluranocene (1) in THF or toluene,

and were found not to decolorize the green solutions within 1 h. However, nitrobenzene (2), 2-methoxynitrobenzene (3), and 1,2-dinitrobenzene (4), even when specially purified, instantly decomposed uranocene solutions and gave brown precipitates. In such experiments the use of substituted uranocenes is more convenient than uranocene itself because of the low solubility of the parent compound. These preliminary studies led to larger scale experiments with nitro compounds. Treatment of 1 with 1 equiv of 2 resulted in the recovery of 85% of the n-butylcyclooctatetraene (5) originally present in the uranocene, and gave a 61% yield of azobenzene (6). A series of substituted nitroarenes was studied, and the results are summarized in Tables I and 11. These yields are probably low owing to handling losses on the small scale used (often less than 100 mg of nitro compound). The recovery of 5 was 85-100% in most of these reactions. Three aliphatic nitro compounds were studied. 2-Methyl2-nitropropane (25) reacts rapidly with 1 to give a mixture of 5 and 1,1,1',1'-tetramethylazoethane (26). Further purification was difficult, since the azo compound is rather volatile. 1

(CH3)$NO2

(CH3)3C-N=N-C(

25

CH3)3

26

1-Nitro-I -phenylpropane (27) gave small amounts of impure azo compound 28 and propiophenone (29). 2-Nitro-20002-7863/78/1500-2433$01 .OO/O

0 1978 American Chemical Society